SUMMARY
TSC2 inactivating mutations elicit mTORC1 hyperactivation and underlie neural and mesenchymal tumorigenesis in tuberous sclerosis complex (TSC). Given the paucity of humanized and lineage-relevant experimental systems relevant to TSC we have developed the first multi-system TSC model, based on CRISPR/Cas9-mediated knockout of TSC2 in human pluripotent stem cells (hPSCs) and their differentiation into putative cell types driving TSC tumors and neurological dysfunction. Mesenchymal neural crest cell (NCC) differentiation recapitulates critical features of lung and kidney tumors, while neural lineage induction of TSC2−/− hPSCs progressively models aberrant neuronal, glial and neural precursor cells (NPCs) that constitute TSC brain tumors. We additionally provide an RNA-sequencing (RNA-seq) resource in which multiple time-points during induction of WT and TSC2−/− cells into neural and neural crest lineages were profiled. Temporal RNA-seq analyses exposed a massive proteostatic stress response activated acutely during neuroepithelial induction in TSC2-deficient cells. Remarkably, this is resolved upon NCC specification but leads to endosomal and metabolic reprogramming in NPCs. Furthermore, TSC2−/− NPCs, but not NCCs, are uniquely vulnerable to death via proteasome inhibition independent of mTORC1 regulation, highlighting cell type-specific proteostatic states. Thus, our data reveal that mechanisms regulating proteostasis stress are strikingly lineage-specific in TSC2−/−cells, and these differences are acquired developmentally. This drives long-term catabolic adaptations specifically in neural cells that counter typical mTORC1 paradigms, resulting in previously unrecognized tumor/lineage-specific therapeutic sensitivities. Overall, our work exemplifies the complexity of elucidating underlying biological mechanisms and therapeutic approaches for multisystem diseases, illustrating the power of utilizing hPSC disease models with tissue-specific relevance.
INTRODUCTION
Inactivating mutations in TSC1 or, more commonly, TSC2 underlie the neoplastic disorders tuberous sclerosis complex (TSC) and lymphangioleiomyomatosis (LAM) (1). TSC is a multisystem disease whereby typically loss of heterozygosity for TSC1 or TSC2 drives development of neurocognitive deficits and low-grade neural and mesenchymal-like tumors in the brain, kidneys, heart, eyes, lungs, and skin (2). Affecting 30% of female TSC patients, and also occurring spontaneously through somatically acquired mutations, LAM takes its form as a progressive pulmonary disease in which clusters of aberrant smooth muscle-like cells expand throughout the lung interstitium, driving the development of tissue-destructive cystic lesions. Although of unknown origin, pulmonary LAM cells are presumed to arise following metastasis of tumorigenic cells from other TSC mesenchymal tumors such as renal angiomyolipoma or uterine tumors, common multisystem features of TSC (3). The age of onset of distinct TSC tumors spans a wide range; for instance, although they persist throughout life cortical tubers and sub-ependymal astrocytomas that develop in the brain arise during embryonic development and can be observed prenatally, while LAM occurs only in post-pubescent females and has a median age of diagnosis of 35 (3, 4). Molecular and functional features are also unique, with neurological TSC tumors expressing markers of immature neuronal differentiation, enhanced gliogenesis, neuronal hyperactivity, and neural stem and progenitor cells (5–7); alternatively, mesenchymal TSC cells display aberrant smooth muscle and adipocyte differentiation (8, 9), increased migratory capacity (10, 11), and expression of female hormone receptors and melanocytic antigens in some populations (12, 13). Thus, the biological consequences of TSC1 or TSC2 deficiency can vary across all stages of development and are likely lineage-dependent. Uncovering the mechanisms driving potential lineage-specific biological responses and adaptations to TSC2 deficiency would contribute greatly to our understanding of the etiology of tumorigenesis in TSC, while also illuminating the mechanisms underlying mTORC1-driven cancers within diverse tissue types.
Acting together within an inhibitory heterocomplex, TSC1 and TSC2 function to negatively regulate mTORC1 signaling. As a dynamic signaling network regulating anabolic and catabolic processes, mTORC1 is a critical regulator of development and tissue homeostasis, whose dysregulation is associated with multiple diseases (14). In TSC, the genetic loss of TSC1 or TSC2 leads to constitutive mTORC1 activation which, depending on the cellular context, can significantly affect cell biology including uncontrolled growth and proliferation, reprogramming of energy metabolism, and altered organelle biogenesis and function (2, 3). Given the tight connection between TSC1 and TSC2-deficiency and mTORC1 hyperactivation, pharmacological mTORC1 inhibition with rapamycin analogues has emerged as a leading clinical strategy to treat many TSC manifestations. Although providing broad clinical benefit, this approach is not curative and produces purely cytostatic relief in responsive patients (15–17); therefore, there is currently a strong focus within the field to identify novel therapeutic strategies to treat, and ultimately eliminate, TSC tumors. Currently available experimental TSC models include TSC1 and TSC2-deficient rodent strains, fibroblasts derived from these animals, and transformed patient tumor-derived cell lines (18–21). Although informative, these tools do not represent the cell lineages or genetic states that are implicated in driving tumor development in TSC patients or, importantly, are not of human origin; thus their capacity to accurately model the multisystem features of this disease is strongly limited. Recently, human PSC-derived TSC1&2-deficient neuronal cultures have been reported which recapitulate the aberrant gliogenesis and hypertrophic neurons observed in TSC brain tumors (22, 23). While demonstrating proof-of-concept, these studies, however, provide limited biological reproducibility and mechanistic insight regarding tumor initiation and progression. Additionally, they lack any evidence and investigation of the aberrant neural precursor populations that persist in patient brain tumors and which likely drive tumor development (5,24,25). A human TSC model that reflects all tumor-relevant cell types and lineages is crucial to truly understand the biology driving lineage-specific manifestations of TSC, and to uncover appropriate therapeutic approaches. Murine models utilizing promoter driven TSC1 and TSC2 knockouts have largely established the cell of origin for neurological TSC tumors as neural stem and progenitor cells (NPCs) (26–28). Although recent human mesenchymal models of TSC do recapitulate multiple aspects of TSC tumors (29), the origin of the mesenchymal features of TSC is less clear. However, much like the neurological component of TSC, evidence suggests mesenchymal-like progenitor cells contribute to tumor formation (2,3,30-32). The mesenchymal progenitor-like qualities of neural crest cells (NCCs) provides the appropriate opportunity to model multiple aspects of mesenchymal TSC tumors in a progenitor cell lineage. Supportive of a NCC model, mesenchymal TSC tumors are broadly composed of mixed cell types that reflect NCC progeny, including adipocytes, smooth muscle cells, and melanocytes, and cutaneous TSC tumors are found in proximity to sites populated by NCCs during development (33). Furthermore, NCCs emerge simultaneously with NPCs, the known origin of TSC brain tumors, at the interface of the neural plate and non-neural ectoderm during neurulation (34).
Here we present a human pluripotent stem cell (hPSC)-based model of the multi-lineage manifestations of TSC. We employed a CRISPR/Cas9 genome engineering approach to introduce inactivating mutations in the TSC2 locus in four well-characterized hPSC lines, allowing us to establish a uniquely robust modeling system permitting analysis of phenotypes driven by TSC2-deficiency among multiple genetic backgrounds. Given the sex-specificity of LAM, two male and two female hPSC lines were included: WA01 (H1, male), WA07 (H7, female), and WA09 (H9, female) embryonic stem cells, as well as 168 1d2 induced hPSCs (168; male, healthy control derived, characterized in (35)). We directed this resource of isogenic stem cells through two parallel neural lineage induction protocols to favor the generation of either NPCs or NCCs. We show that TSC2−/−NPCs and their neuronal and glial progeny, as well as TSC2−/− NCCs, accurately model critical features of neurological and mesenchymal TSC tumors, respectively. Importantly, TSC2−/− NPC cultures reflect distinct cell types that comprise TSC brain tumors, including the aberrant NPC-like population for which experimental models have been particularly elusive. The biological impact of TSC2-deficiency during lineage development, as well as mechanisms of tumor initiation in TSC, are poorly understood; thus, we sought to explore these questions. To this end, we provide a comprehensive resource of RNA-sequencing (RNA-seq) datasets capturing the transcriptional landscape over a developmental time course of neural progenitor and neural crest induction of all four paired wild type (WT) and TSC2−/−hPSC lines. Our analysis surprisingly revealed wide-spread, developmental stage specific, transcriptional alterations in differentiating TSC2−/−cells, and uncovered a large-scale cellular response to an unfolded protein endoplasmic reticulum (ER) stress initiated immediately upon neural lineage induction. Strikingly, we demonstrate that this is selectively resolved in TSC2−/− NCCs but sustained in TSC2−/− NPCs as they are maintained and aged in culture, resulting in long-term alterations in endosomal signaling, proteostasis and metabolic profiles specific to TSC neural tumor cells. Finally, we show that TSC2−/− NPCs are selectively sensitized to proteasome inhibition with clinically relevant compounds irrespective of mTORC1 inhibition with rapamycin. While proteasome targeting has been considered as a broad therapeutic strategy in TSC based on existing experimental models, predominantly as one element of combination therapy, our work suggests that this approach holds promise as a stand-alone therapy or complimentary treatment to existing regimens for the neurological, but not mesenchymal, features of TSC. Collectively, these results exemplify the strength of our multisystem hPSC modeling approach to uncover unappreciated lineage-specific mechanisms that drive and maintain tumorigenic phenotypes and that underlie unique therapeutic vulnerabilities, with high relevance to diverse TSC disease manifestations.
RESULTS
Generation of a TSC2−/− hPSC library using CRISPR/Cas9
To reflect the genetic diversity of patients affected by TSC and permit robust validation of disease-relevant phenotypes, we utilized multiple, well-characterized hPSC lines within this study. As TSC affects both men and women, with LAM affecting exclusively females, two male (H1 & 168 hPSCs) and two female (H7 & H9 hPSCs) lines were selected (Figure 1A). These parental lines were modified at the AAVS1 locus with CRISPR/Cas9 to express mCherry and Zeocin resistance to aid in live cell assays and future in vivo studies (supplementary materials). Due to the higher prevalence and more severe disease phenotypes associated with mutations in TSC2 compared to TSC1 (1), the TSC2 locus was targeted for genetic modification via CRISPR/Cas9. There is a wide array of documented pathogenic mutations in TSC2 (over 2600 unique variants) with no mutational ‘hotspot’ within coding regions or tendency towards a particular mutation type (e.g. small deletion, nonsense mutation) (36–38). To effectively knockout TSC2 expression, exon 3 was selected for targeting due to its ubiquity across all documented variants and its proximity to the N-terminus of its protein product. To achieve the highest possible editing precision across all four TSC2 knockout cell lines, a homology directed repair strategy was employed. Using a single stranded oligonucleotide containing a 35-base ‘stop-codon’ donor sequence (39), a frame shift-inducing insertion containing stop codons in all frames was introduced into TSC2 exon 3, confirmed by PmeI digestion of PCR amplicons of the target site (Figure 1B).
(A) Schematic summary of TSC stem cell modeling strategy. (B) Homologous recombination strategy to knockout TSC2 utilized an ssODN containing a ‘STOP-codon’ donor sequence containing the unique restriction site, PmeI. PmeI digestion of PCR amplicons containing the target cut site reveals homozygous integration of the donor sequence in TSC2−/− hPSC lines. (C) Imaging flow cytometry image capture following staining with TSC2 antibody revealing clonal populations of TSC2−/− cells. (D) Quantification of fluorescence intensity of pluripotency markers OCT4, SOX2, and NANOG in each hPSC line under maintenance conditions, normalized to respective parental WT hPSC lines. Values are the mean ± SEM (n = 12; 3x for each cell line). (E) Haemotoxylin and eosin staining of H9 WT and TSC2−/− teratoma explants. Arrows indicate examples of immature tissues of ectodermal, endodermal, and mesodermal origin. Scale bar, 100µm. (F) Schematic of expected phosphorylation status of mTORC1 effectors under supportive and stress conditions. Western blots probing phosphorylated mTOR (P-mTOR), phosphorylated S6K (P-S6K), and phosphorylated S6 (P-S6) of samples treated for 6h under no treatment, 1% O2, and +100nM rapamycin conditions. Densitometry quantification of western blots displaying the ratio of phosphorylated to total protein of mTOR, S6K, and S6 (n = 8; 2x for each cell line). Statistical significance was determined using two-way ANOVA and Tukey’s post hoc analysis. (G) Schematic representation of monolayer-NPC and EB-NCC differentiation protocols. Red asterisks indicate time points harvested for RNA sequencing.
Homozygous integration of the stop-codon sequence (TSC2−/−) resulted in complete ablation of TSC2 expression, with no detectable protein by immunofluorescence or western blot (Figure 1C, S1A). Much like their WT counterparts, TSC2−/− hPSCs maintain characteristic undifferentiated colony morphology and can be maintained in an undifferentiated state indefinitely and exhibit normal expression levels of hallmark pluripotency markers OCT4, SOX2, and NANOG (Figure 1D), demonstrating that the pluripotency regulatory network functions unperturbed. Furthermore, TSC2-deficiency does not overtly affect pluripotency, as all TSC2−/− hPSCs were able to form tissues of ectodermal, mesodermal, and endodermal origin during in vivo teratoma assays (Figure 1E, S1B). A defining characteristic of both neurological and mesenchymal TSC lesions is their inability to properly regulate mTORC1 signaling, attributed to loss of function of the inhibitory TSC1/2 complex. Under maintenance conditions, both undifferentiated WT and TSC2−/− hPSCs exhibited active mTORC1 signaling as indicated by the phosphorylation of mTOR (Ser2448) and the downstream effectors of the mTORC1 complex, p70 S6K (S6K; Thr389) and S6 ribosomal protein (S6; Ser235/236). However, when exposed to stress conditions (6h at 1% O2), TSC2−/− hPSCs maintain constitutive phosphorylation of the mTORC1 axis, while WT cells can repress this pathway (Figure 1F, S1C). Thus, all four TSC2 knockout hPSC lines successfully recapitulate the mTORC1 molecular signaling abnormalities observed in TSC-associated lesions.
To model both the central nervous system (CNS) and mesenchymal tumors that develop in TSC, we conceived parallel lineage induction approaches to generate cultures enriched for neural precursor cells (NPCs) or mesenchymal-like neural crest cells (NCCs) based on small molecule dual SMAD signaling inhibition (dSMADi) of hPSCs (Figure 1A, 1G). dSMADi promotes rapid exit from pluripotency and induction of a predominant dorsal neuroectoderm (NE) fate, with consequent potential to promote enrichment of NE-derived NPCs or NCCs by defined culture conditions (40, 41). Validating previous findings and our rationale, RNA sequencing (RNA-seq) of end-point cultures revealed that dSMADi of high density adherent monolayer hPSCs resulted in a dominant NPC gene signature (Figure 1G, S1D); conversely, an embryoid body (EB) induction approach permitted NCC marker gene enrichment (Figure 1G, S1D). In this latter protocol, EBs are neuralized for 5 days and then plated onto a fibronectin substrate, which permits outward migration of HNK-1+/SOX9+/p75+ NCCs from PAX6+ NPC-rich cell clusters as they are progressively induced (Figure S1E-G). Thus, we employed the 12-day adherent monolayer and EB-based dSMADi induction strategies (Figure 1G) to respectively establish cultures enriched for NPCs and NCCs from each of our four isogenic paired WT and TSC2−/− hPSC lines. Notably, WT and TSC2−/− end-point cultures similarly expressed appropriate lineage-associated genes (Figure S1D). This included a substantial increase in expression, compared to the undifferentiated state, of NCC determinant genes such as SOX9, PAX3, RXRG, ZEB1, LEF1 and NGFR (encodes p75) uniquely in EB-derived NCC cultures, and a monolayer NPC-specific enrichment of CNS neural progenitor markers including PAX6, SHH, NEUROG2, OTX1, and GLIS3.
TSC2−/− NPCs model key features of TSC neurological manifestations
TSC2-deficiency results in cognitive abnormalities and the growth of low-grade tumors in the developing brain, specifically cortical tubers, subependymal nodules (SENs) and subependymal giant cell astrocytomas (SEGAs), that persist throughout life and contain molecular features indicative of aberrant neural lineage differentiation, maturation, and organelle dysfunction (5). In addition to dysplastic neurons and astrocytes, which to date have been the predominant focus within the field in understanding TSC brain tumors, enlarged cells that demonstrate atypical neural progenitor identity are primary constituents of these tumors (27, 37). To ultimately model both progenitor-like and mature TSC brain tumor cell types, we induced hPSC cultures into NPCs using the high density monolayer dSMADi protocol (Figure 1G). Within 96 hours, we observed that TSC2−/− cells were visibly enlarged and had accumulated an abundance of globular structures which persisted throughout the differentiation time course (Figure S2A). This struck us as being highly reminiscent of vesicle accumulation observed in mTORC1 hyperactive cells in patient tumors (5). WT and TSC2−/− cultures were harvested 12 days following initiation of NPC induction and maintained in neural progenitor expansion medium for multiple passages to examine long-term progenitor phenotypes. TSC2−/− cultures exhibited increased phosphorylation of S6 (P-S6) under fully supplemented media conditions, indicative of mTORC1 hyperactivation (Figure 2A&B, S2B), and were larger than their WT counterparts (Figure S2A&C). Both WT and TSC2−/− cells expressed the NPC lineage markers SOX2, NESTIN and PAX6, demonstrating a progressive increase in expression of NPC determinants and the glial marker GFAP in TSC2−/− cells as they aged in culture (Figure 2A&C, S2D-F). Mitochondrial content was also progressively increased in TSC2−/−NPCs (Figure 2A&D). These phenotypes reflect the aberrant organelle and lineage development observed in patient tumors and imply a dynamic regulation of cell fate and mitochondrial content in NPCs lacking TSC2. Notably, SOX2 expression was increased in TSC2−/− cells not only in the nucleus but also unexpectedly in the cytoplasm, displaying a punctate staining pattern, at later passages (Figure S2G). Additionally, putative vesicle accumulation observed during lineage induction was visibly reduced over the first passage in maintenance cultures but re-emerged progressively over subsequent passages. Altogether these data reveal that TSC2−/− NPC cultures reflect multiple phenotypes reflective of patient brain tumors in TSC and can be exploited to model both disease initiation and progression.
(A) Immunofluorescence images of WT and TSC2−/− NPC maintenance cultures, displaying phenotypic markers of neurological TSC tumors: P-S6, Nestin/PAX6, GFAP/SOX2, and Mitotracker deep red. Scale bar, 50µm. (B) Ratio of P-S6 to total S6 fluorescence intensity in maintenance NPCs, relative to WT levels. Values are ± SEM (n=3; 1x H9, 2x 168 between passages 2-5). (C) Quantification of fluorescence intensity of the lineage markers Nestin (neural; top) and GFAP (glial; bottom) over early, mid, and late passage ranges in culture, relative to WT NPCs. For all relevant figures, passage (p) ranges are defined as: early (p0-2), mid (3–5), late (p6-8). Values are the mean ± SEM (For Nestin: [n=3 early (1 each H7, H9, 168), n=5 mid (2x H1, 1 each H7, H9, 168), n=3 late (1x H1, 2x H7)]; for GFAP: [n=5 early (2x H1, 2x H7, 1x 168), n=4 mid (1x H9, 3x 168), n=3 late (1x H1, 2x H7)]). (D) Relative intensity of Mitotracker deep red staining in NPCs over early, mid, and late passage ranges, normalized to WT NPCs. Values are the mean ± SEM (n=5 early (1x H1, 2x each H9 and 168), n=3 mid (1x H9, 2x 168), n=4 late (2x 168, 1 each H1 and H9) (E) Representative images of WT and TSC2−/− neurons and quantification of the morphological features soma size (left) and neurite outgrowths per cell (right) imaged by endogenous mCherry expression, expressed relative to WT. Values are mean ± SEM (n=3 H9). Scale bar, 50µm. (F) Quantification of the amplitude of AMPA receptor-mediated spontaneous excitatory post-synaptic currents (sEPSCs) measured in H7 and H9 WT (n=6 cells), TSC2−/− (n=7 cells) and TSC2−/− +rapamycin (n=6 cells) neurons in culture. Values are mean ± SEM. (G) Percentage of NCC populations expressing NCC specific lineage markers SOX9 and HNK-1, and NPC specific marker PAX6. Values are mean ± SEM (n = 9; ≥ 2x for each cell line). (H) Quantification of P-S6 immunofluorescence intensity of NCC maintenance cultures exposed to 24h of no treatment, media starvation, and +100nM rapamycin. Values are normalized to no treatment samples within each cell lineage, displaying the mean ± SEM (n = 10; ≥ 2x for each cell line). Statistical significance was determined using two-way ANOVA and Sidak’s post hoc analysis. (I) Mean displacement over 6h of NCC cultures under maintenance conditions evaluated through time lapse imaging (see also supplemental multimedia). Values are the mean ± SEM. Statistical significance was determined using one-way ANOVA and Tukey’s post hoc analysis. (J) Percentage of NCCs reactive to HMB45 immunofluorescence staining. Values are the mean ± SEM (n = 14, ≥ 3x for each cell line). Statistical significance was determined using unpaired t-test. (K) Representative immunofluorescence TSC marker staining of WT and TSCS2−/− NCCs. Scale bar, 50µm. (L) Western blot of maintenance NCCs probing for VEGFD and PDGFRß, quantified via densitometry and normalized to WT protein levels using GAPDH. Values are mean ± SEM (n = 8; 2x for each cell line).
Neuronal differentiation produced TSC2−/−cultures with enlarged soma and an expanded neurite network (Figure 2E) and an increased capacity for gliogenesis in mature cultures (Figure S2H), modeling neuronal structure, network and neural and glial cell fate phenotypes observed in TSC brain lesions. To determine whether TSC2−/− cultures exhibited functional hyperactivation typical of TSC epileptic and cognitive phenotypes, we isolated AMPA receptor (R)-mediated spontaneous excitatory post-synaptic currents (sEPSCs) and measured the amplitude (Figure 2F) and frequency (Figure S2I) of these events via whole-cell patch clamp technique. Indeed, TSC2−/− cells showed significant increases in the amplitude of AMPAR-mediated sEPSCs compared to untreated WT cells. Additionally, this could be rescued by rapamycin during differentiation (Figure 2F), revealing that the neuronal hyperexcitability in TSC2−/− cells is mTORC1-dependent (Figure 2F, S2I). Collectively, our analyses of WT and TSC2−/−progenitor and differentiated neuronal cultures demonstrated that our modelling approach permits the recapitulation of all major cell types observed in TSC brain tumors, reflecting both molecular and functional aspects of these aberrant cells.
TSC2−/− NCCs accurately recapitulate mesenchymal TSC tumor phenotypes
The most significant mesenchymal features affecting the quality of life of TSC patients include renal angiomyolipomas and pulmonary LAM nodules, the latter being associated with high morbidity and mortality (42). These lesions are composed of potential neural crest progeny: angiomyolipomas and LAM nodules consist of proliferative spindle-shaped cells that express smooth muscle markers (e.g., α-SMA) and larger epithelioid cells that react with HMB45 (a monoclonal antibody that recognizes the melanoma-associated protein gp100), with adipocytes contributing to angiomyolipomas only. Other melanocyte markers are also expressed within angiomyolipomas and LAM nodules (e.g., TRYP1, MART1, GD3) (43–45). These data suggest that the cell-type of origin for LAM is neural crest-like. To establish a NCC-based model of mesenchymal TSC, EB-based dSMADi was employed (Figure 1G) to spatially enrich for NCCs (40). By day 10 of differentiation, when NCC specification genes were upregulated in both WT and TSC2−/−cultures (Figure S1D), SOX9+ cells can be observed migrating away from PAX6+ neural clusters (Figure S1E). Selective dissociation of migratory outgrowths at day 12 efficiently enriched for p75 expressing cells in all cell lines, regardless of TSC2 genotype (Figure S1F&G). This resulted in highly enriched SOX9+/HNK-1+ maintenance cultures with minimal contamination of PAX6 expressing NPC populations (Figure 2G, S3A). As expected, P-S6 staining confirms TSC2−/− NCCs are unable to negatively regulate mTORC1 signaling following 24h starvation (Figure 2H). Irrespective of TSC2-deficiency, the generated NCCs are multipotent and can be differentiated to both ectodermal and mesodermal lineages, including smooth muscle cells and adipocytes which are characteristic of TSC mesenchymal tumors (Figure S3B). NCCs also displayed classic mesenchymal stem cell phenotypes. Both WT and TSC2−/−cells displayed expression of epithelial to mesenchymal transition (EMT) related genes during differentiation (Figure S3C); however, TSC2−/− cells showed a significant functional increase in migratory potential compared to WT cells assayed via live cell tracking over a 6h time period (Figure 2I, Supplemental multimedia). In both WT and TSC2−/− NCCs, rapamycin only partially reduced migratory potential of TSC2−/− NCCs.
Common histological markers for mesenchymal TSC tumors include HMB45, α-SMA, and elevated P-S6 indicative of mTORC1 hyperactivation. In addition, PDGFRß expression is elevated in TSC lesions (46–48), and increased serum VEGFD levels are present in TSC patients with angiomyolipomas and LAM (49, 50). To determine whether TSC2−/− NCCs express these hallmark mesenchymal TSC markers, passage 1 NCCs were fixed 24h after completion of the differentiation protocol. Similar to patient tumor explants, HMB45 staining revealed a significant subpopulation of TSC2−/− cells reactive with this marker, displaying similar staining patterns to that seen in primary LAM cells (Figure 2J&K) (51). Interestingly, both WT and TSC2−/− NCCs stained positive for P-S6 and α-SMA at comparable levels (Figure 2K, S3D). In addition, immunoblotting of NCC maintenance cultures revealed expression of the TSC biomarker VEGFD in both WT and TSC2−/−NCCs, with increased PDGFRß protein levels in TSC2−/− NCCs (Figure 2L). The expression of multiple hallmark TSC tumor markers in both WT and TSC2−/− cells suggests that NCCs are inherently well suited to model mesenchymal TSC. However, elevated HMB45 reactivity and PDGFRß expression, together with the enhanced migratory mesenchymal phenotype of TSC2−/− NCCs, strongly support the use of these cells as an accurate model of the mesenchymal manifestations of TSC and LAM.
TSC2-deficiency drives large-scale cell type-specific transcriptional dysregulation upon NPC and NCC lineage induction
We next sought to elucidate the molecular basis for the development of tumors and broad disease phenotypes in TSC2−/−NPC and NCC progenitors. Given that cell lineage induction is heavily determined by dynamics in transcriptional regulatory programs, we reasoned that elucidating the genome-wide transcriptional events underlying the generation of these cell types would offer critical insights and reveal whether developmental genetic programs underlie tumor development in TSC. Thus, we performed genome-wide RNA-seq along a developmental time course of TSC2−/− and WT hPSCs as they were induced to differentiate through the NPC and NCC lineages using our parallel dSMADi strategies (Figure 1G). RNA-seq timepoints were selected to monitor global gene expression throughout the high-density monolayer NPC-differentiation protocol corresponding to states of pluripotency (day 0), early neural induction (day 1), neuralization (days 3&5), and neural precursor specification and maturation (days 8&12). To identify transcriptional changes at various stages of the EB-based NCC protocol, RNA-seq time points were similarly selected to emphasize phases of transition during differentiation: EB-formation (day 1), neuralization (days 2&4), NCC specification (day 7) and enrichment (day 10) (Figure 1G). RNA-seq was performed on all four isogenic paired lines, serving as biological replicates for analysis of expression patterns within each respective differentiation time course for a total of 96 sequenced libraries. Principle component analysis (PCA) revealed distinct trajectories between WT and TSC2−/− samples within each differentiation time course, indicating dynamic transcriptional changes between genotypes from pluripotency to target cell lineage (Figure 3A). Clear separation between WT and TSC2−/− samples was attributed to principle component (PC) 2 during NPC differentiation and to PC4 during NCC induction, implying that TSC2-deficiency is more impactful to transcriptional variance in NPC differentiation.
(A) Representative three-dimensional plotting of PCA of monolayer-NPC (H1 & 168) and EB-NCC (H7 & H9) differentiations. (B) Number of total DEGs identified (FDR < 5%) over the RNA-seq differentiation time-course in monolayer-NPC and EB-NCC RNA-seq datasets. (C) Log2 fold change (Log2FC) of hallmark neuroectodermal (NE) genes over the monolayer-NPC RNA-seq time-course. (D) Log2FC of hallmark NCC specification genes over the EB-NCC RNA-seq time-course. (E) Comparative gene ontology (GO)-Biological Function enrichment analysis of upregulated DEGs (padj < 0.05) between matched TSC2−/− and WT samples at each respective time point of monolayer-NPC and EB-NCC differentiations. Red box highlights enrichment of stress response GO terms.
Confirming our findings that TSC2-deficiency does not affect maintenance and pluripotency of hPSCs (Figure 1D&E), the number of differentially expressed genes (DEGs) between WT and TSC2−/− cells at day 0 (pluripotent state) was negligible (6 and 19 genes identified respectively in NPC and NCC data sets) (Figure 3B). Likewise, very few DEGs were detected at day 1 of NCC induction (59 genes), a time-point representing a change in cell substrate interactions (switch from Matrigel adherence to suspension EB formation) but prior to the initiation of neuralization using dSMADi. Strikingly, TSC2−/− cells in both differentiation paradigms demonstrated a substantial number of DEGs compared to WT within 24 hours of neural induction (day 1 NPCs, 621 genes; day 2 NCCs, 189 genes), which expanded to thousands of DEGs at 3 days post-induction (day 3 NPCs, 3626 genes; day 4 NCCs, 2830 genes) (Figure 3B). While the TSC2-mTORC1 signaling axis has been implicated in transcriptional regulation in various biological settings this has not been previously evaluated in the context of lineage induction during embryonic development. The magnitude of DEGs that we observed within 24-72 hours of neural induction was highly unexpected and indicates that TSC2 is a potent regulator of genome-wide transcriptional programs during NE fate determination, potentially affecting NPC and NCC specification.
During both induction protocols, the number of WT versus TSC2−/− DEGs peaked 3 days following the addition of neuralization cues and remained substantially elevated at each respective end-point (day 12 NPCs, 898 genes; day 10 NCCs, 609 genes) (Figure 3B). Notably, NPC cultures exhibited larger gene expression differences across all differentiation time-points compared to NCCs. Confirming that our parallel differentiation strategies generate distinct cell populations, 42-57% of DEGs at end-point compared to day 0 in WT cultures are unique to their target cell type (Figure S4A). In TSC2−/− cultures, cell-type specific DEGs at endpoint increased up to 65% (Figure S4A). Together, this reveals that while there is a large degree of overlapping DEGs between NPC and NCC-induced cultures, as is expected given their similar developmental origins, NPC and NCC-induced cultures also display distinct transcriptional profiles reflecting their unique cell lineages and the differential phenotypes in patient neurological versus mesenchymal lesions.
To evaluate the influence of TSC2-deficiency on specific gene expression profiles, differential gene expression was evaluated between adjacent differentiation time points of NPC and NCC differentiation. Genes displaying significant fold-change in at least one set of adjacent time points were selected and grouped by similarity of expression trajectories into multiple clusters using DP_GP_cluster (52). The largest gene clusters within WT and TSC2−/− datasets of respective NPC and NCC analyses were selected and found to display similar trajectories with a high degree of overlap in DEG membership (Figure S4B). Indeed, at differentiation end-point, TSC2−/−cells did not display significantly altered expression of NE and NCC genes in NPC and NCC cultures, respectively, compared to WT samples (Figure S1D). Upon further investigation of gene sets relevant to NE and NCC formation, we did observe a reduction in NE gene expression (Figure 3C) in TSC2−/− cells during NPC differentiation, most prominently during mid-stages of differentiation. No change in NCC gene signatures was observed during NCC differentiation (Figure 3D). Thus, while TSC2-deficiency delays NPC fate induction, it does not overtly impede differentiation towards either the NPC or NCC lineage.
To assess the biological context of TSC2-deficiency on differential gene expression across NPC and NCC differentiation, gene ontology (GO) enrichment analysis was performed on significant DEGs between TSC2−/− and WT samples at each time point. The most dominant GO biological process terms resulting from this analysis were related to endoplasmic reticulum (ER)/proteostatic-stress, response to unfolded protein, and autophagy (Figure 3E, red box). During TSC2−/− NPC differentiation these GO term groupings were over-represented consistently across most timepoints, all of which were significantly enriched until day 5, with autophagy-related terms being maintained until day 8. In contrast, these terms were only over-represented at days 2 and 4 of TSC2−/− NCC differentiation. This temporal enrichment pattern is consistent with KEGG pathway analysis, which revealed enrichment for “protein processing in the ER”, “lysosome”, “phagosome”, and “antigen processing and presentation” in TSC2−/− compared to WT cultures (Figure S4C). Similar to GO enrichment analysis (Fig 3E), the KEGG pathway terms associated with ER/proteostatic stress response and catabolic vesicular signaling are enriched only transiently during TSC2−/− NCC differentiation, specifically at days 4 and 7. Of note, these time points are associated with the neuralization stages of differentiation (days 2 and 4) and precede NCC enrichment (day 7) when adherent neuralized clusters predominate the differentiating cultures. This implies that the activation and stability of an ER/proteostatic stress and catabolic signaling response in TSC2−/− cells is associated with neuralization of the differentiating cultures and maintenance of an NPC fate. Corroborating this, expression of NE determinant genes in both WT and TSC−/− cultures was largely induced at days 3 and 5 in monolayer-NPC cultures, further increasing or remaining high through day 12. These same genes were induced to a similar degree at day 4 (and some at day 7) in EB-based NCC cultures but did not increase further or were completely lost at later time points (Figure S4D). In contrast, expression of neural crest lineage genes was much higher in EB-based NCC cultures, initiating at day 4 and further increasing through day 10, revealing the progressive emergence of a dominant NCC fate in these cultures following initial neuralization (Figure S4E).
TSC2−/− NPCs and NCCs exhibit unique catabolic signalling profiles during development
RNA-seq analyses of NPC and NCC differentiation revealed significant enrichment of genes involved in lysosomal and autophagy signaling in differentiating TSC2−/− populations. These genes were differentially expressed throughout the entire monolayer-NPC differentiation protocol, whereas during EB-based NCC induction, these enrichment terms were transient. To investigate this further, hierarchical clustering was performed on all RNA-seq samples comparing TSC2−/−to WT gene expression at each time point, focusing on genes associated with lysosome signaling and formation, along with positive and negative regulation of autophagy (Figure 4A-C). As expected, NCC differentiation sample day 4, the height of EB neuralization, clusters with NPC differentiation days 3 and 5, showing distinctive upregulation of lysosome-associated genes (Figure 4A, red box). In contrast, early NPC time points cluster well with both early and late NCC differentiation time points and display limited differential gene expression from WT cells (Figure 4A, green box). Interestingly, NCC day 4 and NPC days 3 and 5 display similar expression patterns of genes associated with negative regulation of autophagy (Figure 4B, red box); however, NPC days 3 and 5 cluster separately from NCC day 4 when profiled with genes positively regulating autophagy (Figure 4C, red box). In the latter case, NCC day 4 continues to cluster with NPC time points, separate from all other NCC samples (Figure 4C, yellow box). Thus, gene-set specific hierarchical clustering of DEGs between WT and TSC2−/− cells in both NPC and NCC datasets reveals that lysosome/autophagy gene expression dysregulation reaches its peak at differentiation time points associated with neuralization of the respective cell populations, and that this catabolic signaling phenotype persists uniquely within the NPC lineage. Lysosome staining of differentiating NPC cultures with both a LAMP1 antibody and Lysosensor live cell dye confirmed that lysosomal content is significantly increased in TSC2−/− cells throughout the differentiation time course post-neuralization (Figure 4D&E, S5A). This phenotype is rescued when the mTORC1 inhibitor rapamycin is included in culture medium starting 6 hours prior to dSMADi, revealing the dependency of TSC2 on mTORC1 for lysosomal regulation during early neural lineage induction. As this is counter-intuitive, considering how mTORC1 activation is typically thought to repress catabolic signaling, we investigated the sub-cellular localization of the lysosomal transcription factor TFE3, which is typically excluded from the nucleus upon mTORC1 activation resulting in inhibition of lysosomal biogenesis. Corroborating our findings, TFE3 nuclear localization was increased in TSC2−/− NPCs, in a rapamycin-dependent manner, with the largest changes occurring during days 3 and 5 of differentiation (Figure 4F, S5B), when catabolic gene expression was most highly dysregulated in these cultures in our RNA-seq data sets.
(A) Heatmap featuring hallmark lysosome genes displaying Log2FC of TSC2−/− compared to WT samples at each respective timepoint of monolayer-NPC and EB-NCC RNA-seq datasets. Red box highlights time points associated with increased lysosome gene expression. Green box highlights time points of minimal differential lysosome gene expression. (B&C) Heatmaps of genes involved in negative and positive regulation of autophagy, displaying Log2FC of TSC2−/− compared to WT samples at each respective time point of monolayer-NPC and EB-NCC RNA-seq datasets. Red boxes highlight timepoints associated with increased differential gene expression. Yellow box highlights clustering of NCC d04 time point with late NPC timepoints featuring moderate differential gene expression. (D) Representative images (at differentiation day 5) and quantification of LAMP1 immunofluorescence intensity (mean spot analysis) in the presence and absence of rapamycin throughout monolayer-NPC differentiation. Values are mean ± SEM (n = 10; 1x H9, 3x H1, 2x H7, 4x 168). Scale bar, 50µm. (E) Quantification of the cytoplasmic area of Lysosensor staining relative to WT at day 5 of monolayer-NPC differentiation. Values are mean ± SEM (n = 10; 1x H9, 3x H1, 2x H7, 4x 168). (F) Representative images and quantification of fluorescence intensity relative to WT of nuclear TFE3 signal at day 5 of monolayer-NPC differentiation. Values are mean ± SEM (n = 10; 1x H9, 3x H1, 2x H7, 4x 168). Scale bar, 50µm. (G) Immunofluorescent staining of attached neuralized EB clusters and migratory NCC outgrowths at day 7 EB-NCC differentiation. Dotted lines delineate the boundary of attached neuralized EB clusters. Scale bar, 100µm.
Our findings suggest that NCCs resolve this developmental catabolic response as they are specified and acquire migratory capacity. To investigate this, day 7 EB differentiation cultures were fixed and probed for lysosome and autophagosome indicators by immunostaining and paired with lineage specific markers for NCCs (SOX9) and NPCs (PAX6) (Figure 4G, S5C). As expected, neuralized rosette clusters stain positive for PAX6, while migrating NCCs are SOX9+. Neuralized rosette clusters, however, display a clear increase in lysosomal content compared to SOX9+ cells, as indicated by LAMP1 expression and Lysosensor staining (Figure 4G, S5D). Furthermore, autophagosome content was also increased in neural clusters as indicated by LC3β staining and autophagosome staining with Cyto-ID live cell dye (Figure 4G, S5E). Of note, lysosomal and autophagosome content appeared increased in TSC2−/− neuralized EB clusters compared to WT cells. Both LAMP1 and LC3β were detected in migrating SOX9+ cells; however, their expression is visibly diminished compared to NPC-rich clusters. Mirroring hierarchical clustering observations of ‘positive’ and ‘negative’ regulators of autophagy (Figure 4B&C), TSC2−/− monolayer cultures demonstrated reduced autophagosome staining compared to isogenic WT cultures during early induction. This is an expected consequence of TSC2 loss, however, an atypical increase in autophagosome content was observed at late differentiation (Figure S6A). This suggests a dynamic regulation of autophagy signaling in TSC2−/− cells particularly as they differentiate through the neural lineage, with a lineage-dependent induction of autophagy signaling which, together with their increased lysosomal content, reflects the observed accumulation of vesicular structures in these cultures (Figure S2A) as well as TSC tumors specifically within the brain (5, 53).
TSC2-deficiency drives long-term catabolic and proteostatic dysregulation in NPCs but not NCCs
To determine whether the observed developmental cell stress responses we observed led to long-term lineage-specific differences in catabolic signaling mechanisms, we measured levels of endosomal vesicles in NCCs and monolayer culture-induced NPCs after multiple passages in maintenance culture. Reflecting developmental phenotypes, lysosomal content was highly increased in TSC2−/− NPCs (Figure 5A) compared to WT counterparts, but only slightly elevated in NCCs (1.4-fold versus 6-fold in NPCs) (Figure 5B&C). Rapamycin treatment had no effect on Lysosensor levels in either cell type; thus, unlike during lineage development lysosomal content can no longer be normalized with short-term mTORC1 inhibition. Autophagosome content was unchanged between WT and TSC2−/−NCCs (Figure 5C) as measured by CytoID live cell dye; however, we observed a 7-fold increase in autophagosome intensity in TSC2−/−NPCs compared to WTs (Figure 5D). Autophagosome intensity was further increased following chloroquine treatment, demonstrating that both WT and TSC2−/− NPCs undergo autophagic flux (Figure 5D). This indicates that increased autophagosome content in TSC2−/− NPCs is primarily a consequence of increased biogenesis of these vesicles and not flux inhibition, contrary to the current paradigm of how mTORC1 hyperactivation is presumed to direct autophagy signaling. This unexpected observation was corroborated by a progressive increase in LC3β and autophagosome levels during TSC2−/− NPC induction (Figure 5E, S6A), revealing an early establishment of this altered autophagy mechanism during lineage development.
(A) Representative fluorescence imaging for high content analysis of Lysosensor stain and LAMP2A immunofluorescence in WT and TSC2−/− maintenance NPCs; quantification of Lysosensor intensity with and without 100nM rapamycin, relative to WT NPCs. Values are mean ± SEM (n = 7, representing all cell lines). Scale bar, 50µm. (B) Representative fluorescence images of maintenance WT and TSC2−/− NCCs; quantification of Lysosensor signal relative to WT NCCs. Values are mean ± SEM (n = 9; ≥ 2x for each line). Scale bar, 50µm. (C) Quantification of autophagosome (Cyto-ID staining) fluorescence intensity in maintenance NCCs, relative to WT NCCs. Values are mean ± SEM (n = 9; ≥ 2x for each line). (D) Representative images for high content analysis of Cyto-ID live cell staining; quantification of autophagosome fluorescence intensity in maintenance NPCs, relative to WT NPCs, with and without 5-20nM chloroquine (CQ). Values are mean ± SEM (n = 13 for basal analysis, n = 7 (WT) and 8 (TSC2−/−) for +CQ samples, representing all cell lines). Scale bar, 50µm. (E) Representative Western blot in WT and TSC2−/− cultures during monolayer-NPC differentiation with and without 20nM rapamycin initiating 6h prior to differentiation. Proteins assessed are TSC2, ribosomal protein S6 and P-S6 at Ser-235/246, ULK1 and P-ULK1 at Serine-555 (AMPK-dependent), and LC3ß. (F) Quantification of Proteostat staining during monolayer-NPC differentiation with and without 100nM rapamycin. Values are mean ± SEM (n=10; 1x H9, 3x H1, 2x H7, 4x 168). (G) Representative images of proteostat aggresome staining in WT at day 5 monolayer-NPC differentiation and day 7 of EB-NCC differentiation. Scale bar, 50µm, 100µm.
To investigate likely mechanisms by which TSC2−/−NPCs are capable of promoting autophagy in the presence of hyperactive mTORC1 activity, we examined the phosphorylation status of ULK1, a kinase with essential roles in the early stages of pre-autophagosome formation. Phosphorylation of ULK1 at serine-757 is a critical event through which mTORC1 inhibits autophagy signaling. We confirmed both mTORC1 hyperactivation, based on increased P-S6 (serine-235/236) (Figure 5E, S6B), and increased phosphorylation of ULK1 serine-757 in TSC2−/− cells during NPC induction (Figure S6D). AMPK is a secondary kinase with ULK1-dependent activity, with its phosphorylation of serine-555 driving autophagy. While AMPK levels were similar across genotypes (Figure S6E), we observed an increase in ULK1 serine-555 phosphorylation in differentiating TSC2−/−NPCs (Figure 5E, S6F), revealing activation of this alternative autophagy-promoting pathway in these cells. Corroborating pro-autophagy signaling in TSC2−/− NPCs, total, unprocessed and lipidated forms of LC3β were increased in TSC2−/− cultures during monolayer-NPC induction, indicating increased autophagosome content (Figure 5E, S6G). An increased LC3β-II/ LC3β-I ratio was also observed, demonstrating that autophagic flux is also elevated in TSC2−/− cells during NPC induction (Figure S6G).
Comparative gene ontology (GO) enrichment analysis of our RNA-seq data (Figure 3E) illustrated an early activation of the misfolded protein response, which may function as a driving event for catabolic signaling activation in TSC2−/− cells during differentiation. We confirmed an increase in protein aggregate accumulation in TSC2−/− cells during neural differentiation using Proteostat aggresome indicator dye (Figure 5F). This phenotype was transient, however, with aggregate levels peaking at day 8 of differentiation and reducing to normal levels in end-point TSC2−/−NPC cultures. Reflecting endosomal phenotypes and lineage specificity in proteostasis regulation, Proteostat intensity was more predominant in differentiating neural clusters than in migratory NCCs (Figure 5G). Accordingly, we observed low levels of aggregates during early stages of maintenance culture, but by late passages reflecting extensive aging they had accumulated selectively in TSC2−/− NPCs (Figure S6H). Together these data demonstrate that TSC2−/−cells mount an early developmental response to increased protein production and suggest that TSC2−/− NPCs maintain selective reliance on lysosome-autophagy signaling to limit protein aggregate accumulation over time.
Underscoring the concept of an early protein stress response in TSC2−/− cells, levels of the mTORC1 effector P-S6K and especially its target P-S6, a mark of active protein translation, were strongly reduced within the first 24 hours of dual SMAD inhibition in WT cultures, but this inhibition was incomplete at early time points in TSC2−/− cells (Figure 5E, S6B&E). Interestingly, P-S6 is undetectable by day 8 of NPC differentiation in spite of mTORC1 hyperactivation in TSC2−/− cells (Figure 5E). P-S6 staining of day 7 NCC differentiations reveal P-S6+ migratory NCCs delaminating from largely P-S6 negative neural clusters (Figure S6C), revealing that NPCs specifically down-regulate S6 phosphorylation during lineage induction. Demonstrating the dependence of this signaling axis on mTORC1 regulation, P-S6 levels were strongly reduced by rapamycin treatment (Figure 5E); partial restoration of P-ULK1 serine-555, LC3β and autophagosome levels by rapamycin reflects a mechanistic connection between proteostasis and endosomal regulation in TSC2−/− NPCs (Figure 5E, S6A).
The bioenergetic profile of TSC2−/−cells is lineage-specific
mTORC1 is a well-established regulator of cellular metabolism, and it is becoming increasingly clear that endosomal signaling is highly integrated with cellular bioenergetics. Previous cell models have revealed metabolic abnormalities in TSC1 or TSC2-deficient cells and have largely focused on increased glycolytic capacity. However, given the pervasive cell type-specific differences in endosomal signaling observed in our system, we sought to investigate the lineage-specific metabolic profiles of TSC2−/−cells. TSC2−/− NPCs accumulate mitochondrial content in our model (Figure 2A&C), as well as in patient tumors (5) and exhibit elevated levels of reactive oxygen species (Figure S7A), indicating likely effects on mitochondrial metabolism and oxidative phosphorylation in neural cells. Extracellular flux analysis at resting levels and under conditions of induced mitochondrial stress revealed that TSC2−/−cells possess an overall increased capacity for mitochondrial oxygen consumption as NPCs (Figure 6A) but that this effect is limited in NCCs (Figure 6B; S7B&C). Increased extracellular acidification in these same assays in TSC2−/− NPCs (Figure 6C), but not NCCs (Figure 6D), additionally indicate that TSC2−/− NPCs uniquely exhibit increased glycolytic flux. Both TSC2-deficient NPCs and NCCs have a higher capacity for ATP production compared to WT cells (Figure 6E&F). In NPCs, this is attributed to both glycolytic and oxidative pathways (Figure 6E); in NCCs, this is due solely to oxidative capacity and is observed only under fully stressed conditions and not at resting state (Figure 6F).
(A&B) Plots showing the O2 consumption rate at each time-point in mitochondrial stress test extracellular flux assays for NPC (A) and NCC (B) cultures. Measurements 1-3 were readings taken at resting state; 4-6, following addition of the complex V inhibitor Oligomycin (Oligo); 7-9, after addition of the mitochondrial membrane potential uncoupler FCCP; and 10-12, after addition of the complex I and III inhibitors Rotenone (Rot.) and Antimycin A (AA). Values are mean ± SEM (NPCs: n = 7 (2x H1, 3x H7, 1x H9, 1x 168); NCCs: n=7 (1x H1, 2x each H7, H9, 168)). (C&D) Plots showing the extracellular acidification rate (ECAR) at each time-point in mitochondrial stress test, with time-points and treatments as described for panels 6A&B. In addition, the plasma membrane Na+/K+-ATPase agent Monensin (Mon.) was added, to stimulate maximal ATP demand and glycolytic capacity, and measured at # 13-15. Values are mean ± SEM, with replicates as in panels 6A&B. (E&F) ATP production rates (JATP) from glycolysis (left-hand bars), and Oxidative Phosphorylation (OxPhos) (right-hand bars) at resting and maximal levels in NPCs (E) and NCCs (F). Values are mean ± SD, with replicates as in panels 6A&B. (G&H) Integration of JATP from OxPhos and glycolysis reveals a high bioenergetic capacity of TSC2−/−NPCs between resting and maximal (‘max’) states (G), while bioenergetic capacity of WT NPCs (G) and WT and TSC2−/− NCCs (H) is minimal. Values are mean ± SD, with replicates as in panels 6A&B. (I) Spare respiratory capacity, representing the difference in O2 consumption at measurements 7-9 compared to 1-3, in NPCs (left) and NCCs (right). Values are mean ± SD, with replicates as in panels 6A&B.
These data suggest a high degree of bioenergetic flexibility specifically in TSC2-deficient NPCs that permits these cells to uniquely adapt to metabolic stress. To directly test this hypothesis, we measured the amount of ATP produced in each cell type from oxidative phosphorylation and glycolysis throughout the mitochondrial stress test using a recently published data analysis platform (54). This integrated analysis confirmed that TSC2−/− NPCs exhibit a high bioenergetic capacity compared to their WT counterparts (Figure 6G), capable of increasing ATP generation under conditions of metabolic stress (‘max’) compared to ‘resting’ states, while TSC2-deficient NCCs exhibit limited metabolic flexibility (Figure 6H). Comparing oxygen consumption post-treatment with the mitochondrial oxidative phosphorylation uncoupler FCCP with the resting state additionally demonstrates that TSC2-deficiency imparts an increased spare respiratory capacity to a much larger degree in NPCs than in NCCs (Figure 6I). Altogether, our data demonstrate that TSC2-deficient cells adopt substantial catabolic and metabolic signaling adaptations in a highly lineage-specific fashion that correlates with the relative nature of developmental proteotoxic stress responses.
Clinically-relevant proteasome inhibitors permit selective, lineage-specific cytotoxicity of TSC2−/−cells
The accumulation of lysosomes and autophagosomes, paired with the abundance of aggresomes within NPC populations as they both differentiate and self-renew is indicative of the importance of functional proteolytic machinery in the maintenance of cell homeostasis in this cell lineage. TSC2-deficiency exacerbates these phenomena, offering a potential vulnerability of TSC2−/− cells that could be exploited for therapeutic benefit if protein degradation pathways are rendered inoperative pharmacologically. To test this, both NPCs and NCCs were exposed to the FDA-approved proteasome inhibitors bortezomib and carfilzomib, as well as the lysosomal inhibitor chloroquine, both alone and in combination. To investigate the compatibility and/or potential synergistic effects of these treatments with mTOR inhibitors, these drugs were administered both in the presence and absence of rapamycin. After 48h exposure to proteasome inhibitors, selective toxicity to TSC2−/− NPCs was observed at 40nM, with minimal effect on WT NPCs (Figure 7A&B). At 200nM of bortezomib or carfilzomib, TSC2−/− NPC samples continued to display increased sensitivity to proteasome inhibition, although WT cells began to show low level toxicity. These trends were observed following 24h exposure to proteasome inhibitors, though less pronounced (Figure S7D&E). In contrast, both WT and TSC2−/− NCCs showed nearly identical toxicity profiles with increasing concentrations of proteasome inhibitors at both 24h and 48h (Figure 7A&B, S7D&E). Interestingly, treatment with the mTORC1 inhibitor rapamycin or targeting the lysosomal/autophagy pathway with chloroquine did not affect the viability of either NPCs or NCCs (Figure 7C, S7F), nor did chloroquine or rapamycin display any synergistic cytotoxic effects when combined with bortezomib or carfilzomib (Figure S7G). Taken together, these data demonstrate that TSC2−/− NPCs, but not WT NPCs, are specifically sensitized to proteasome inhibition. Furthermore, these effects are cell-lineage dependent, as both WT and TSC2−/− NCCs are not able to tolerate low levels of proteasome inhibition (Figure 7D).
(A-C) Toxicity observed in NPCs (left) and NCCs (right) 48 hours after treatments with Bortezomib (A), Carfilzomib (B) and Rapamycin or Chloroquine (C). (D) Schematic representation of ER stress states and resulting responses to proteosome inhibition. Values are the mean ± SEM. (A-B) n=7 for NPCs; 1x for H7, 4x for H9 and 2x for 168. n=11 for NCCs; 2x for H1, 2x for H7, 3x for H9 and 4x for 168. (C) n=4 for NPCs; 1x for H7and 168, 2x for H9. n=6 for NCCs; 1x for H7 and H9, 2x for H1 and 168. Statistical significance (p < 0.05) was established by Two-way ANOVA and Tukey’s post hoc analysis.
DISCUSSION
This study presents the first multi-lineage model of the neoplastic and neurological disorder TSC and provides a rich resource of the transcriptomic signatures that accompany human NCC and NPC development in both WT and TSC2−/− conditions. These data have implications for understanding mechanisms of mTORC1-driven tumorigenesis more broadly, and, in the case of congenital malignancies, demonstrate how a single mutation within a ubiquitous pathway can manifest itself in a tissue specific manner. Specifically, our modeling system permits analyses of the initiation and progression of multiple disease cell types and aberrant phenotypes relevant to TSC, including neuro-cognitive dysfunction and neural and mesenchymal tumorigenesis, importantly using a human stem cell model. Recently reported TSC stem cell models have focused on investigating aberrant glial and neuronal cells (22, 23); however, these cell types are non-proliferative components of neural TSC lesions and thus the consequences of TSC2-deficiency in the precursor cell pools that drive tumorigenesis were overlooked in these studies. Given that TSC1−/−and TSC2−/− mouse models die embryonically with failed neural tube closure (55–57), and that proliferative neural and mesenchymal progenitor cells have been implicated as cells of origin for TSC tumors (2), we reasoned that loss of TSC2 in NPCs and mesenchymal-like NCCs would have substantial biological consequences reflective of disease etiology. Our system offers an unprecedented opportunity to investigate the development and maturation of both TSC2−/− NPCs and NCCs, which we now show broadly reflect the mesenchymal tumors that develop in multiple tissue systems such as kidney (angiomyolipoma) and lung (LAM) in TSC patients. By employing genome-wide RNA-seq and molecular approaches using multiple genetic backgrounds, we demonstrate that the early development of TSC2−/−NPCs and NCCs is associated with aberrant activation of a proteostatic stress response, which is differentially resolved in neural and neural crest precursors. This leads to long-term molecular adaptations in neural cells that endow lineage-specific differences in catabolic activity that become independent of mTORC1 signaling.
Our findings have critical implications for understanding the pathological basis of tumor development in TSC, and for the treatment of system-specific disease manifestations. Several biological mechanisms have been proposed to contribute to tumorigenesis in TSC, many of which have undergone investigation as potential therapeutic targets. Dysregulation of autophagy has been clearly established as a component to tumor formation (58); however, lysosome and autophagosome accumulation has been reported only in TSC brain tumors and in neural-lineage specific murine and cell models of TSC (5, 53). Our data reconcile these previous observations by revealing that TSC2−/− neural lineage cells, but not NCCs or hPSCs, uniquely develop elevated endosomal signaling mechanisms, and that these are established during early stages of precursor cell specification. Mitochondrial and metabolic defects, as well as proteasome dysfunction, have also been identified as potential therapeutic targets for both neurological and mesenchymal TSC (59–64). These previous studies did not, however, permit analysis of the development of these phenotypes during lineage specification; thus, it has been entirely unclear which biological mechanisms drive the induction of disease phenotypes, and which are secondary adaptations of TSC2−/− cells. Our findings reveal for the first time that proteostatic stress is a driving event in the developmental origin of TSC-relevant tumor cell types, and that lineage-specific signaling responses result in the unique metabolic and endosomal adaptations that define long-term biological differences of neural and mesenchymal TSC tumors.
The lineage specific catabolic adaptations we identified in TSC2−/− cells pose consequences in the therapeutic context. Bortezomib and carfilzomib are FDA-approved proteasome inhibitors that have been considered for treatment of TSC manifestations, and this has been applied predominantly toward mesenchymal tumors. Strikingly, we show that TSC2−/− NPCs can be selectively targeted for cell death using these agents, leaving WT NPCs unaffected. However, the same treatments display no genotype specific toxicity in NCCs. These latter findings are in contrast to previous studies demonstrating the potential therapeutic effectiveness of these treatments in vitro; however, we note that the TSC2-deficient cell types used in these studies were murine cell lines with limited tissue-specific relevance to TSC tumors (65–67). Indeed, in a preclinical trial using a Tsc2 heterozygous knockout mouse model, bortezomib treatment was ineffective in targeting mesenchymal renal tumors (68). Our data, which utilizes four unique human genetic backgrounds, supports this conclusion and additionally demonstrates that the utility of proteasome inhibition therapy for TSC is limited to the neural lineage manifestations. These results underscore the importance of using stem cell-based models to generate relevant cell types when investigating mechanisms of disease using in vitro systems, particularly in the case of multisystem neoplastic disorders.
Our findings reveal a decisive role for translational regulation and catabolic signaling following neural lineage induction from pluripotency. Following the addition of neural induction cues, levels of P-S6, a marker of active protein translation, decrease sharply in both WT cells and TSC2−/− cells in spite of hyperactive mTORC1 signaling in TSC2−/− cells. Additionally, we establish that AMPK/ULK1-mediated pro-autophagy signaling is concurrently activated in differentiating TSC2−/− cells, corroborating previous observations of mTORC1-independent regulation of autophagy in patient tumors and TSC1/TSC2-deficient cell lines (53). Our data introduce a paradigm in which decreased global protein translation is favored under normal conditions during early neural cell fate commitment, and that TSC2−/− NPCs, which cannot regulate mTORC1-mediated protein translation and degradation (69), potently activate catabolic signaling mechanisms as an adaptive response to promote proteolysis. We further demonstrate that while these lineage-specific metabolic adaptations are reversed by rapamycin treatment initially during lineage induction they ultimately become rapamycin insensitive, consistent with a progressive reliance on mTORC1-independent signaling mechanisms to maintain endosomal signaling in these cells. This biology is reflected in the lack of synergistic toxicity when rapamycin is combined with proteasome inhibitors to selectively target TSC2−/− cells. However, this catabolic adaptive response is resolved upon NCC specification, revealing dynamic, lineage-specific regulation of endosomal, metabolic and proteostasis signaling. Although lysosomal-autophagy signaling is intrinsically active in WT NPCs, this system is exacerbated and cumulative in TSC2−/− cells. While maintenance TSC2−/− NPC cultures exhibit intact autophagic flux at a level comparable to WT cells, we show that they exhibit an overall increase in the biogenesis of endosomal vesicles which display incomplete clearance and progressive accumulation over time. The fact that proteosome inhibitors, but not autophagic flux inhibition (chloroquine), are selectively toxic to TSC2−/− NPCs clearly demonstrates that these cells are ultimately dependent on proteolytic activity to maintain their survival. The increased accumulation of protein aggregates in late stage TSC2−/− NPC cultures, a time-point at which their growth is dramatically inhibited, further supports this conclusion.
It is apparent that tight regulation of catabolic signaling is integral in the development of neural lineages. Indeed, catabolic signaling is quickly revealing itself to be an important pathway across multiple levels of neural development, with involvement in the exit from pluripotency, lineage specification, and the maintenance and activation of adult neural stem cell populations (70–72). Here, we advance these previous observations to show that this phenomenon is highly specific to the neural lineage, with the developmentally related neural crest lineage adopting distinct mechanisms. Additionally, although several studies have investigated the metabolic consequences of TSC2-deficiency in neural and glial lineages, there is limited information regarding derivation of these phenotypes in the neural progenitor pools that gives rise to TSC brain tumors, and sparse knowledge of mesenchymal progenitor pools contributing to TSC. The comprehensive RNA-seq dataset presented here compiles the transcriptomic profiles of early NPC and NCC differentiation, shedding light on the developmental origin of TSC neoplasms while presenting insight into the role of TSC2-mTORC1 signaling in emerging progenitor populations and their contribution to neoplastic disease. This resource and in vitro model system are together intended to serve as a platform to further investigate the mechanisms underlying TSC tumorigenesis that remain elusive, such as the sex-specific mesenchymal manifestations of LAM, and to identify previously unconsidered vulnerabilities that span multiple TSC2-deficient cell lineages.
Author Contributions
Conceptualization, S.P.D., L.M.J., and W.L.S.
Methodology, S.P.D. and L.M.J.
Investigation, S.P.D., L.M.J., A.P., J.Y-L., C.D., T.T.W., V.C.D., A.R., D.A.P., M.E.H., and H.S.
Formal Analysis, S.P.D., L.M.J., A.P., J.Y-L, T.T.W., and S.H.
Writing – Original Draft, S.P.D. and L.M.J.
Writing – Review & Editing, S.P.D., L.M.J., A.P., J.Y-L., D.A.P., M.E.H., H.S., and W.L.S.
Funding Acquisition, S.P.D., L.M.J., and W.L.S
Resources, W.L.S.
Visualization, S.P.D., L.M.J., J.Y-L.
Supervision, W.L.S.
Declaration of Interests
The authors declare no competing interests.
MATERIALS AND METHODS
Cell Culture
hESC lines H9 (WiCell WA-09), H7 (WiCell WA-07), H1 (WiCell WA-01) and induced pluripotent stem cell line 168 (35) were maintained on diluted Matrigel (BD Biosciences #354230) in Essential 8 (E8) media: DMEM/F-12 (Life Technologies #11330), 0.064µg/ml Ascorbic Acid 2 Phosphate Mg (Sigma #A8960), 14µg/ml sodium selenium (Sigma #S526), 0.1µg/ml FGF2 (Life Technologies #PHG0263), 19.4µg/ml human Insulin (Wisent #511-016-CM), 10.7µg/ml transferrin (Sigma #T0665), 0.002µg/ml TGF-β1 (Life Technologies #PHG9202), 543µg/ml NaHCO3 (Sigma #5761) (73). Media was replaced daily, and PSCs were passaged every 4-5 days using 0.5mM EDTA.
NCCs were maintained on poly-L-ornithine (PLO; 0.0015%)/fibronectin (10µg/ml) coated growth surface in neural induction media (NIM): 1:1 DMEM/F-12: Neural Basal Media (Life Technologies #21103-049), 0.5x N2 Supplement (Life Technologies #17502-048), 0.5x B27 Supplement (Life technologies #17504-044), 5µg/ml human Insulin, 0.02µg/ml FGF2, 0.02µg/ml hEGF (Sigma #E9644), 0.5x GlutaMax (Life Technologies #35050061). Media was replenished every second day. For passaging, 90% confluent NCCs were dissociated with Accutase and plated at a density of 20,000-25,000 cells/cm2, approximately every 4-6 days, until a maximum of 10 passages. NPCs were maintained on diluted Matrigel in NIM (Matrigel diluted 1/18 in DMEM-F12 for coating). For passaging, cells were dissociated with Accutase and plated at a density of 100,000-150,000 cells/cm2, at the highest end of this range for the first 2 passages. Cells were supplemented with 10µM y-27632 (ROCKi; Tocris #1254) for the first passage 2h prior to harvest and for 48h post-plating. A half-media change was performed every 2-3 days with 75% of media replenished weekly. Cells were passaged when confluent, typically 2-3 weeks for the first 2 passages, and 1-1.5 weeks thereafter for a maximum of 10 passages, when growth of TSC2−/−NPCs markedly slows. All cells were maintained at 37°C, 5% O2, 10% CO2.
CRISPR/Cas9 Genome Editing
At 80% confluence, hPSCs were dissociated with Accutase (STEMCELL Technologies #07920). Cells were rinsed off of the growth surface and gently triturated. Accutase was then diluted 1:4 in PBS−/− (Life Technologies #14190-250) and then pelleted (250g, 5 min). Cells were resuspended in PBS−/− for counting. For each CRISPR/Cas9 transfection sample, 1 × 106 cells were aliquoted into microcentrifuge tubes and pelleted (250g, 5 min). Pellets were resuspended in 100µl electrolyte buffer E2 (Thermo Fisher #MPK10096) with 3µg CRISPR/sgRNA plasmid (Addgene #48139) (74) and 1.75µg of TSC2 exon 3 ssODN donor sequence (Table 1; Integrated DNA Technologies) or 3µg AAVS1:mCherry donor plasmid (supplemental information). Samples were electroporated using the Neon Transfection System (Thermo Fisher MPK5000) using 100µl tips (Thermo Fisher #MPK10096) at 2 pulses of 1050v, 30ms pulse width. Electroporated cells were then plated onto diluted Matrigel at low density in E8 media containing 10µM ROCKi. Electroporated cultures were left to grow until individual colonies formed. Expanding colonies were picked into Matrigel coated 96-well cluster plates for PCR/restriction enzyme-based screening for edited cells. Karyotypic analysis revealed no chromosomal abnormalities in 7 of the 8 CRISPR/Cas9 edited cell lines utilized in this study (data not shown). Trisomy 12, a common chromosomal abnormality in hPSC cultures (75), was detected in a subpopulation of H1 TSC2−/− cells (35%) and is, thus, not related to CRISPR/Cas9 editing. Guide RNA (gRNA) sequences utilized to modify the TSC2 and AAVS1 loci (Table 1) were identified and high scoring (98 and 80, respectively) (76) gRNA sequences were selected to minimize potential off target cleavage events. To identify potential CRISPR/Cas9-induced off-target mutations in TSC2−/− hPSCs, the highest scoring predicted guide RNA off-target sites were identified using the CRISPOR algorithm (77). Sequencing of these regions revealed no off-target cleavage induced mutations in any of the TSC2−/− hPSC lines. See table 1 and supplementary methods for DNA/plasmid/oligonucleotide sequences.
(GE: genome editing, ssODN: single stranded oligonucleotide, PCR: polymerase chain reaction)
PCR Screening/Genotyping of CRISPR/Cas9 Edited Cells
Genomic DNA was harvested using QuickExtract solution (screening of CRISPR/Cas9 clones; Epicentre EPI-QE0905T) or NucleoSpin Tissue kit (routine genotyping; Macherey-Nagel #740952.25). OneTaq DNA polymerase (New England BioLabs #M0480) was utilized using primers flanking the genome editing target site of exon 3 of TSC2. Cycling parameters are as follows: 94°C initial denaturation for 30s, 25x [94°C 20s, 56.5°C 30s, 68°C 30s], 68°C 5m, 4°C ∞. PCR amplicons were then digested with PmeI restriction enzyme (New England BioLabs #R0560S) in OneTaq buffer for 1h at 37°C. Agarose gel electrophoresis was utilized to resolve digested amplicons indicating successful integration of ‘stop codon’ donor sequence.
EB-NCC Differentiation
Day 0: Aggrewell plates (STEMCELL Technologies #34425, #34415) were prepared using pluronic solution (2% pluronic w/v in sterile H2O), centrifuging at 1,300g for 5min, followed by 2 washes with prewarmed DMEM/F-12. Half differentiation culture volume of NIM +10µM ROCKi was added to each well and placed at 37°C until ready for use. One day prior to routine passaging, 10µM ROCKi was added to pristine undifferentiated hPSCs 2h before dissociation with Accutase. Cells were incubated in Accutase for 10-20min, until cells were easily washed off of the growth surface. Cell suspension was diluted 1:4 in DMEM/F-12 and pelleted at 250g for 5min. Cells were resuspended in NIM +10µM ROCKi and counted. Appropriate volumes of cell suspension were added to Aggrewells to generate 750cell EBs, and wells were topped up to 5ml (6 well format) or 2ml (24 well format) with NIM+ROCKi. Aggrewell plates were centrifuged at 100g for 3min and placed at 37°C, 5% O2, 10% CO2. Day 1: 24h later, all media was carefully removed from each well to minimize disturbing newly formed EBs in microwells. NIM +10µM SB431542 (Tocris: 1614) and 500nM LDN-193189 (Stemgent: 04-0074) (NIM+SB/LDN) was carefully added to each well and Aggrewell plates were returned to 37°C, 5% O2, 10% CO2. Days 2-4: Half media changes with NIM+SB/LDN. Day 5: EBs were dislodged from microwells using a P1000 and transferred to 15ml conical tubes to allow EBs to gravity settle. Media was refreshed with NIM+SB/LDN, and EBs were plated onto PLO/fibronectin coated plates at a density of 40 EBs/cm2. EBs were placed at 37°C, 5% O2, 10% CO2 and dispersed evenly over the growth surface. Days 6-11: Media was replaced every other day with NIM+SB/LDN. Day 12: Neuralized adherent EB clusters were carefully washed off of the growth surface using a P1000, and media was aspirated. Cells were dissociated with Accutase and passed through a 40µM mesh filter to remove any remaining cell clumps. Accutase was diluted 1:4 with DMEM/F-12, and enriched NCCs were pelleted (250g, 5min), resuspended in NIM, and seeded onto PLO/fibronectin coated culture ware at a density of 25,000-30,000 cells/cm2.
NCC Neural/Glial/Mesenchymal Differentiation
At routine passaging, maintenance NCCs were seeded at a density of 20,000-25,000 cells/cm2 onto PLO/fibronectin coated tissue culture ware in NIM. Differentiation towards neural, glial, and mesenchymal fates was initiated 24h later. Neural: NIM was replaced with basal NIM (bNIM; NIM without insulin, hEGF, and FGF2) supplemented with 50ng/µl BMP2 (STEMCELL Technologies #78135). Glial: NIM was replaced with bNIM supplemented with 20ng/ml Heregulin-β1 (STEMCELL Technologies #78071) for 5 days. Media was then changed to bNIM supplemented with 5µM Forskolin (STEMCELL Technologies #72112) and 10% FBS (Life Tech #12483-020). Mesenchymal: Adipocyte differentiation was carried out using the MesenCult Human Adipogenic Differentiation kit (STEMCELL Technologies #05412) according to the manufacturer’s instructions. Smooth muscle cell (SMC) differentiation was carried out by replacing NIM with bNIM supplemented with 10ng/ml TGF-β3 (Sigma #SRP3171). After 72h, media was replaced with 1:1 blend of bNIM and Media 231 (Life Technologies #M-231-500) supplemented with smooth muscle growth supplement (SMGS; Life Technologies #S00725) and 10ng/ml TGF-β3. After a total of 7 days, media was replaced with Media 231 supplemented with SMGS and 10ng/ml TGF-β3. Differentiating SMCs were passaged at 90-95% confluence when needed. All differentiations were carried out for 14 days, with media replacement every other day. Cultures were fixed with 3.7% PFA at endpoint for analysis. Oil red staining of adipocytes was carried out as previously described (78).
Monolayer-NPC Differentiation
hPSCs were passaged as intact colonies, maintained on Matrigel, and neural lineage differentiation was initiated when the cultures were 90%+ confluent to promote neuroepithelial induction (41), typically 3-5 days post-passage. Neuroectoderm differentiation was initiated by replacing E8 with KSR media containing 10µM SB and 500nM LDN. KSR media (for 500mL) contains 414mL of Knockout-DMEM (Life Tech, #10829-018) with 15% Knockout-serum replacement (Life Tech, #10828-028), 5mL of 100X Glutamax (Thermo Fisher, #35050061), 5mL of 100X MEM-NEAA (Gibco, #11140-050), 500uL α-ME (Thermo Fisher, #21985023), and 500µL Gentimicin (Wisent, #450-135). Media was changed daily as KSR+ SB & LDN from days 0-3. At day 4, N2 media was added in increasing amounts in the following ratios: days 0-3 100% KSR, days 4-5 75:25 KSR:N2, days 6-7 50:50 KSR:N2, days 8-9 25:75 KSR:N2, days 10-11 100% N2, harvest at day 12. LDN supplementation was maintained at all days, while SB was included only on days 0-3. N2 media composition is as follows (for 500mL): 486.5mL of DMEM-F12, 15% glucose, 5mL N2 supplement, 20µg/mL human insulin, 2.5mL Hepes (Life Tech, #15630080), and 500µL Gentimicin. Cells were maintained at 37°C, 5% O2, 10% CO2 during differentiation.
NPC-neuronal Differentiation
hPSC-derived NPCs at passage 0-2 were harvested with Accutase and plated on PLO/Laminin-coated glass coverslips (Thermo Fisher, #12-545-101) at a density of 10,000-20,000 cells/cm2. For substrate coating, coverslips were first sterilized in EtOH and washed extensively in sterile 1X PBS−/−, then placed in plastic wells and submerged in PLO (Sigma, P4957) solution (15% of 0.01% stock in sterile H20) for 2h at room temperature or overnight at 4°C. PLO was aspirated and immediately replaced with 25µg/mL laminin in 1X PBS−/−, incubated at 37°C for 2h and stored at 4°C for up to 1 week or at −20°C for long-term storage. Differentiation to neuronal cultures was performed as described (23) with some modifications. Briefly, differentiation was induced when cells were ∼70% confluent, typically 2-3 days post-plating, by replacing media with Neuronal differentiation media, containing 50:50 DMEM-F12: Neurobasal, 2.5mL N2 supplement and 5mL B27 supplement per 500mL, and 20ng/mL each of BDNF (Peprotech, 450-02) and GDNF (Peprotech, 450-10), 500µg/mL Dibutyryl cAMP (Sigma, #D0627), and 200nM L-ascorbic acid (Sigma). Cells were also transferred to normoxic conditions at the time of differentiation induction (37°C, 5% CO2). Cells were fed twice weekly by replacing 75% of spent with fresh media. Cells were differentiated for 4-6 weeks prior to collection for functional assays or fixation for IHC. Cell fixation was performed by removing all but 250µL of cell media from each well (24-well format) and adding 250µL of 8% PFA, to limit loss of fragile neurite networks. Cells were fixed on ice for 1h, and then at room temperature for 30 minutes before carefully removing PFA, washing gently once in 1X PBS−/− and storing at 4°C in 1X PBS−/−.
Whole-cell Patch Clamp Recording
Cells were collected from glass coverslips in 24-well culture plates. Before transfer to the recording chamber in an upright Nikon Eclipse E600FN microscope, culture medium was gradually changed to oxygenated artificial cerebral spinal fluid (ACSF) containing (mM): 124 NaCl, 5 KCl, 1.25 NaH2PO4, 1.2 MgSO4, 26 NaHCO3, 2 CaCl2, and 10 glucose, in 30 mins at room temperature (79). Whole-cell patch clamp recordings were obtained from differentiated neurons at 4-6 weeks following the initiation of the differentiation procedure (80). Patch electrodes with a resistance of 5-10 M were prepared from borosilicate glass capillaries with a Narishige micropipette puller (Model PC-100, Tokyo, Japan). Pipette intracellular solution contains (mM): 130 K-Gluconate, 2 MgCl2, 0.6 EGTA, 10 HEPES, 5 KCl, 2 ATP-Mg(Na2), pH 7.3. Data were collected using MultiClamp 700B amplifier. Signals were filtered at 2kHz, digitized at 20kHz by a Digidata 1500 interface, acquired by the pClamp 10.7 software, and analyzed with the Clampfit 10.7 program (Molecular Devices). To characterize the intrinsic membrane properties of cells, hyper- and depolarizing rectangular pulses of 500-1000ms duration were applied. Spontaneous EPSCs (sEPSCs) were recorded at a −60mV holding potential for at least 10 mins to isolate AMPAR mediated currents from NMDAR mediated currents. Additionally, picrotoxin (PTX) (100M), a GABAA blocker, was administered to further pharmacologically isolate AMPAR mediated currents. sEPSC events were detected automatically using Clampfit 10.7, and frequency and amplitude histograms were constructed using this program. All events were confirmed visually on the basis of the rise and decay times. All electrophysiology data were expressed as mean ± SEM. Data was first tested for normality using the Shapiro-Wilk normality test. Then, statistical significance was assessed using a one-way ANOVA test followed by post hoc Bonferroni test, and a Mann-Whitney rank test for comparisons of data that were not normally distributed. Statistical significance differences are established as p<0.05.
Teratoma Assays
Teratoma assays were performed as previously described (81). Briefly, hPSCs were harvested using Accutase 1 day prior to routine passaging, diluted 1:4 in DMEM/F-12, and pelleted (250g, 5min). Cells were resuspended in E8 media and counted. 1 × 106 cells were aliquoted, pelleted, and resuspended in 100µl ice cold Matrigel for each planned injection. Matrigel embedded cells were injected intramuscularly into the tibialis anterior of NOD/SCID mice. Once large visible leg tumors were observed (8-12 weeks), mice were sacrificed and teratomas were excised and fixed in 10% formalin. Teratomas were then processed for paraffin embedding, sectioned, and stained with hematoxylin and eosin for analysis.
Imaging Flow Cytometry
hPSCs were dissociated with Accutase, diluted 1:4 in PBS−/−, and then pelleted (250g, 5min). Cell pellets were resuspended in 3.7% PFA and fixed for 15min at room temperature. Fixed cells were washed twice with excess PBS−/−, followed by permeabilization in ice-cold 90% methanol for 30min on ice, followed by 2 washes in staining buffer: 1% BSA (Wisent # 800-095-EG), 0.1% Triton-X (BioShop #TRX777) in PBS−/−. Samples of 5 × 105 cell were aliquoted into 5ml round bottom tubes and blocked in excess staining buffer for 30min on ice. Cells were pelleted and resuspended in 100µl of 1:100 TSC2 antibody (Table 2) diluted in staining buffer and incubated at room temperature for 1h. Cells were washed in excess blocking/staining buffer, and resuspended in 100µl of 1:1000 secondary antibody diluted in staining buffer and incubated for 30min at room temperature, shielded from light. Cells were washed once with excess staining buffer and resuspended in 200µl containing 2µg/ml Hoechst 33342 (Invitrogen #H3570) diluted in PBS−/−. Single stain and no stain controls were prepared in parallel. Cells were passed through a 40µm cell strainer (Falcon #352340) prior to imaging on the Amnis ImageStreamX Imaging Flow Cytometer.
(IF = Immunofluorescence, IFC = Imaging flow cytometry, FC = flow cytometry, WB = western blot)
Flow Cytometry
EB-NCC cultures were rinsed with media to remove neuralized adherent EB-clusters using a P1000, prior to harvesting with Accutase. Matching undifferentiated hPSC lines were harvested in parallel with Accutase. Cells were gently lifted via trituration, diluted 1:4 in DMEM/F-12, and passed through a 40µm cell strainer. Cells were pelleted, resuspended in NIM, counted, and 5 × 105 cells were aliquoted into round bottom tubes for staining. Cells were blocked in excess FACs buffer (4% FBS [Life Tech #12483-020], 100µM EDTA in PBS−/−) for 15min on ice. Cells were pelleted and resuspended in 100µl 1:20 diluted p75-Alexa Fluor 647 conjugated antibody (BD Biosciences: #560326) in FACS buffer. Cells were incubated for 30min on ice, shielded from light. Cells were washed once with excess FACS buffer and resuspended in 500µl of 1:2000 diluted Sytox Blue nucleic acid stain (Thermo Fisher #S11348) in FACS buffer. Mouse IgG1-Alexa Fluor 647 isotype controls (BD Biosciences #557783) and no stain controls were prepared in parallel. Cells were passed through a 40µm filter and immediately measured using the BD Biosciences LSR-Fortessa flow cytometer and analyzed with FloJo (version 10.5.3).
Western Blotting
Cell lysates were collected after washing cells with ice cold PBS−/−, followed by brief incubation on ice with RIPA cell lysis buffer: 150mM NaCl, 1% Triton-X, 0.1% SDS, 50mM Tris, 1x PhosSTOP phosphatase inhibitor mix (Sigma # 4906845001), 1x complete protease inhibitor mix (Sigma 11873580001). Cells were scraped and transferred to microcentrifuge tubes and rotated at 4°C for 30min, followed by centrifugation at 16,000g at 4°C for 20min to generate cleared lysates. Lysates were stored at −80°C. After thawing for analysis, protein concentrations were normalized and probed via western blot, imaged using the Odyssey Classic Infrared System (LI-COR Biosciences), and analyzed using standard densitometry techniques using FIJI (82). Antibodies utilized for western blot can be seen in Table 2.
Live Cell Staining
NPC or NCCs were dissociated on day of routine passaging and seeded into 96-well cluster plates for live high content imaging assays. NPCs were seeded at a density of 20,000 cells per well and NCCs were seeded at a density of 7,500 cells per well. 24h after plating media was replenished and, if applicable, cells were administered treatments of rapamycin or chloroquine. After 24h incubation (or 72h for post-rapamycin Lysosensor staining of NPCs), cells were prepared for live imaging by washing with media, followed by the addition of live cell dyes in NIM with 2µg/ml Hoechst 33342: 1:1000 Lysosensor Green DND-189 (Thermo Fisher #7535), 1:2000 Cyto-ID Autophagy Detection dye (ENZO # ENZ-51031), 20nM Mitotracker Deep Red FM (Thermo Fisher #M22426), or 25µM carboxy-H2DCFDA (Thermo Fisher #I36007). Cells were incubated with live cell dyes for 30min under maintenance conditions. Cultures were then carefully washed three times for 5min with NIM, and media was replenished with appropriate treatments for imaging. Imaging for high content quantification was performed using the Thermo Fisher ArrayScan VTI High Content Screening (HCS) platform, utilizing the live cell chamber at 37°C, 5% CO2. Image analysis was facilitated by HCS Studio software (version 6.5.0).
Cytotoxicity Assays
Cells were seeded and grown following maintenance conditions for 24 hours. They were then treated with 100nM Rapamycin (Calbiochem #553211) and/or 10µM Chloroquine (Sigma, C6628) and/or various concentrations of Bortezomib (Adooq Bioscience #A10160) or Carfilzomib (Adooq Bioscience #A11278) as well as 0.1% DMSO for 24 or 48 hours. They were then stained with 2µg/mL Hoechst and 10nM Sytox Green (Thermo Fisher #S7020) for 30 minutes. Imaging was performed using the Thermo Fisher ArrayScan VTI HCS platform, utilizing the live cell chamber at 37°C, 5% CO2. Image analysis was facilitated by HCS Studio software, in which cell death was assessed using the percentage of Sytox Green positive cells in a given population.
Immunofluorescence
Unless specified for specific antibody targets, all immunostaining was performed as follows: Cells were fixed with 3.7% PFA for 15min at room temperature after removing media, followed by 3 washes with PBS−/−. Cells were permeabilized and blocked in excess immunofluorescence staining buffer (IF-SB; 1% BSA, 0.1% Triton-X in PBS−/−) for 1h at room temperature. IF-SB used for blocking was removed, and IF-SB containing diluted primary antibodies (see Table 2) was added to each sample and incubated overnight at 4°C. Samples were then washed three times for 10min before the addition of secondary antibody diluted 1:1000 (Table 2) in IF-SB. Samples were incubated in secondary antibody solution for 1h at room temperature, followed by three washes with PBS−/−, in which 2µg/ml Hoechst 33342 was included in the final wash step. Images were captured using the Zeiss Observer A1 or Observer Z1 fluorescence microscopes or captured for high content analysis using the ArrayScan VTI HCS platform and Studio software.
Alterations to the immunostaining protocol: HNK-1 cell surface staining was completed after PFA fixation, but prior to cell permeabilization. For HMB45 staining, samples were permeabilized using 100% ice-cold methanol for 10min on ice following PFA fixation. For LAMP1 and LC3ß immunostaining, cells were fixed in ice-cold 100% methanol prior to proceeding with blocking and staining.
Proteostat Staining
Cells were fixed with 3.7% PFA for 15min at room temperature, followed by three washes with PBS−/−. Cells were then permeabilized with 0.1% Triton-X diluted in PBS−/− for 10min, followed by three washes with PBS−/−. Proteostat aggresomes detection dye (ENZO #ENZ-51035-K100) was diluted 1:2000 in PBS−/− containing 2µg/ml Hoechst 33342. Samples were incubated for 30min at room temperature, followed by three washes, 10min each, with PBS−/−. Images were captured using the Zeiss Observer A1 or Observer Z1 fluorescence microscopes, or captured for high content analysis using the ArrayScan VTI HCS platform and Studio software.
Motility Assay
NCCs were dissociated for plating on day of routine passaging and seeded into 96-well cluster plates at increasing densities of 1,000 to 6,000 cells per well in NIM. 24h later, media was replenished and 100nM rapamycin or DMSO was added to appropriate samples. Time lapse imaging (15 minute intervals over 6h) of mCherry or CellTracker Deep Red (Thermo Fisher #C34565) signal was performed utilizing the Thermo Fisher ArrayScan VTI HCS platform, utilizing the live cell chamber at 37°C, 5% CO2. CellTracker Deep Red staining was performed according to the manufacturers recommended protocol. Cell tracking image analysis was performed on wells displaying 30-60% confluence, using HCS Studio software.
Extracellular Flux Analysis
Extracellular flux assays were performed by Seahorse technologies, using the 96XF platform (Agilent Technologies), as previously described for mitochondrial stress test (29). Modifications are: cells were seeded in Seahorse XF plates on Matrigel at 50,000 cells/well for NPCs and 25,000 cells/well for NCCs 1 day prior to analysis. Monensin was used, at 20µM, to stimulate maximal ATP demand and glycolytic capacity through the plasma membrane Na+/K+-ATPase (83). Each ‘measurement’ recorded consisted of a 3-minute mixing period at resting state or following addition of indicated chemicals, followed by a 3-minute interval during which O2 consumption rates were actively measured and recorded. O2 consumption rates were normalized to the number of cells plated, based on an equivalent # of cells remaining in WT and TSC2−/− wells at assay end-point (Figure S7B&C). ATP production rates and Bioenergetic capacity were calculated as described (54).
RNA-Seq sample preparation and sequencing
Cells were dissociated using Accutase and diluted 1:4 in PBS−/− prior to pelleting (250g, 5min). Cells were resuspended in PBS−/− and total cell count was noted. Cells were pelleted (250g, 5min) and resuspended in TRIzol reagent (Thermo Fisher # 15596026). Samples were stored immediately at −80°C until processing for RNA extraction. RNA extraction was performed in batches including all differentiation time points within a WT and TSC2−/− cell line pairing. RNA extraction was performed according to the manufacturer’s protocol followed by DNase digestion (Qiagen #79254) and ethanol precipitation of RNA. RNA concentration was estimated using NanoDrop 2000 spectrophotometer (Thermo Fisher ND-2000) and ERCC spike in (Thermo Fisher #4456740) was added, normalizing to cell number. Quality control of RNA integrity and total RNA concentration was then measured using the Bioanalyzer Eukaryote Total RNA Pico kit (Agilent #5067-1513). Only samples with RIN values of 8.0 or greater were submitted for sequencing. Total RNA H9 WT and TSC2−/− NCC sample libraries were prepared with ribosome depletion (Illumina # 20020598), all other sample libraries were prepared using the TruSeq Stranded mRNA kit (Illumina #RS-122-2101). All samples were sequenced on the Illumina NextSeq500 or NovaSeq6000 platform (Illumina) generating at least 30 million reads per sample. Reads were assigned to transcripts from GENCODE release 27 using Salmon (84) and differentially expressed genes were identified using DESeq2 v1.20.0 (85). DESeq2 analysis modelled fold change between time points using the cell line as an independent factor (using the model ‘∼cell_line + condition’), and fold changes were calculated using lfcShrink function, applying the apeglm method (v 1.2.1) (86). Gene ontology enrichment analysis was performed using clusterProfiler R package (87). Time course gene expression clustering was performed using Dirichlet process Gaussian process mixture model using the DP_GP_Cluster Python package (52) to cluster TPM values for genes that show a significant expression difference (using a 5% FDR threshold) between sequential timepoints. Separate gene lists were calculated for NPC and NCC time courses, and expression patterns were clustered separately for WT and TSC2−/−cells within each protocol. All RNA sequencing datasets are available upon request and will be submitted to GEO.
Statistical analysis
Sample data acquired from H9, H7, H1 and 168 hPSCs and their derivatives were considered biological replicates within a given assay or experiment. Multiple biological replicates from each cell line were utilized within a given experiment whenever possible. When data were normalized to WT samples, test sample data were normalized to their respective isogenic WT control before determining mean values, etc. Unless otherwise indicated, statistical significance was assessed using a one-way ANOVA test followed by Bonferroni post hoc analysis. When only 2 groups were compared a student’s t-test was used.
SUPPLEMENTAL FIGURES
(A) Western blot probing for TSC2 in WT and TSC2−/−hPSC lines. (B) Haemotoxylin and eosin staining of H7, H1, and 168 WT and TSC2−/− teratoma explants. Arrows indicate examples of immature tissues of ectodermal, endodermal, and mesodermal origin. Scale bars, 100µm. (C) Western blots probing for total mTOR, S6K, and S6 of samples treated for 6h under no treatment, 1% O2, and +100nM rapamycin conditions. (D) Heatmap displaying log2 fold change of select hallmark NPC and NCC genes at RNAseq endpoint for monolayer-NPC and EB-NCC differentiations. (E) Representative immunofluorescence staining of lineage specific markers PAX6 (neural ectoderm), SOX9 and HNK-1 (neural crest) in WT and TSC2−/− cultures at d12 EB-NCC differentiation. Scale bar, 100µm. (F) Representative flow cytometry analysis of p75 expression in WT and TSC2−/− cultures at d12 EB-NCC differentiation prior establishing maintenance populations. (G) Percent p75+ cells at d12 of EB-NCC differentiation. Error values are the mean ± SEM (total n = 8; 2x for each cell line).
(A) Phase images of WT and TSC2−/−cells during monolayer-NPC induction at differentiation days 5 and 12, displaying increased cell size and vesicle accumulation in TSC2−/− cultures. Scale bar, 50µm. (B) Exemplary western blot of H1 and H7 NPC cultures, probing for TSC2, P-S6 and S6, revealing constitutive S6 phosphorylation in TSC2−/− cells. (C) Quantification of NPC volume, relative to WT by Cellomics ArrayScan imaging analysis. Values are the mean ± SEM (n=13, including all cell lines). (D) Quantification of PAX6 fluorescence intensity in NPCs, relative to WT NPCs. Values are mean ± SEM (n = 5; 1x H7, 2x for H1 and 168 at p0-3). (E) Quantification of SOX2 fluorescence intensity in NPCs over early, mid, and late passage ranges, normalized to WT NPCs. Values are mean ± SEM (n = 6 early, n = 5 mid, n = 3 late). (F) Immunofluorescence image of SOX2 staining in TSC2−/−NPC cultures. (n = 4 early (2x 168, 1x each H7 and H9), n = 9 mid (2x H1, 3x H7, 1x H9, 3x 168), n = 3 late (1x H1, 2x H7)). (G) Quantification of SOX2 subcellular localization, displaying the ratio of TSC2−/− to WT fluorescence intensity within nuclei (left) or cytoplasm (right). (H) Representative immunostaining for neuronal marker MAP2 (top image) and glial marker GFAP (bottom image). Graph (bottom panel) displaying % of cells in neuronal cultures expressing high levels of GFAP relative to WT. (n = 4; 2x each H7 and H9). Scale bar, 50µm. (I) Representative traces of current clamp recordings of action potentials (left) and voltage clamp recordings of AMPA receptor-mediated sEPCs (top) in WT, TSC2−/−, and TSC2−/− +rapamycin neurons. Quantification of sEPSC frequency (bottom). Values are mean ± SEM.
(A) Representative immunofluorescence staining of passage 1 maintenance NCC cultures, probing for neural crest specific markers, SOX9 and HNK-1, and the neuroepithelial marker, PAX6 (n = 8; 2x for each line). Scale bar, 100µm. (B) Representative immunofluorescence staining of WT and TSC2−/−168 NCCs differentiated towards neural (neuronal markers: Tuj1, Nestin), glial (Schwann cells: GFAP, maintaining SOX10 expression at endpoint) and mesenchymal lineages (smooth muscle cells: Calponin; adipocytes: oil red staining). Scale bars, 50µm. (C) Expression of hallmark EMT genes at d10 of EB-NCC differentiation. (D) Quantification of immunofluorescence intensity of mesenchymal TSC markers, P-S6 (n = 10; 3x for H1, H7, 168, 1x for H9) and α-smooth muscle actin (n = 12; 3x for each line), in maintenance NCCs.
(A) Venn diagrams featuring the intersections of significantly up/down regulated genes of WT and TSC2−/− cultures at differentiation endpoint. Genes included were those with FDR<0.05, and a log2FC of >1.0 (up-regulated) or >-1.0 (down-regulated) at end-point compared to day0 of the given cell type. (B) Representative gene clusters displaying similar expression trajectories in WT and TSC2−/−cultures throughout monolayer-NPC and EB-NCC differentiations. (C) Comparative KEGG pathway enrichment analysis of upregulated DEGs (padj < 0.05) in TSC2−/− samples compared to WT at each respective time point of monolayer-NPC and EB-NCC differentiations. Red boxes highlight enrichment of stress response pathways. (D) Expression profiles of neuroepithelial genes throughout NPC and NCC differentiation, with the gene set analyzed indicated at the top of each graph. (E) Expression profiles of neural crest specification genes throughout NPC and NCC differentiation, with the gene set analyzed indicated at the top of each graph.
(A) Quantification of fluorescence intensity of Lysosensor live cell stain in WT and TSC2−/− cultures during monolayer-NPC differentiation, ± 20nM rapamycin. Values are mean ± SEM (n=10; 1x H9, 3x H1, 2x H7, 4x 168); representative images of Lysosensor staining of differentiating NPCs at day 8 of differentiation. Scale bar, 50µm. (B) Representative images of TFE3 immunofluorescence staining of monolayer-NPC cultures at differentiation day 5 in H1 and H7 lines, ± 20nM rapamycin. Scale bar, 50µm. (C) Example immunofluorescence staining of neuralized EB clusters (PAX6+) and migratory NCCs (SOX9+) at day 7 of EB-NCC differentiation. Dashed lines delineate borders of attached neuralized EB clusters. Scale bar, 100µm. (D&E) Example Lysosensor (D) and Autophagosome (E) live cell staining at day 7 EB-NCC differentiation. Dashed lines define the borders of adherent neuralized EB clusters. Scale bars, 100µm.
(A) Quantification of CytoID autophagosome intensity in monolayer-NPC cultures at day 1 and 8 of lineage induction, in the absence and presence of 20nM rapamycin. Values are mean ± SEM (n=10; 1x H9, 3x H1, 2x H7, 4x 168). Scale bar, 100µm. (B) Quantification by densitometry of 1 representative Western blot, as shown in Figure 5E, per cell line (H1, H9, 168). Results displayed are log2FC of TSC2−/− compared to WT. (C) Representative P-S6 immunofluorescence staining at day 7 of EB-NCC differentiation cultures. Dashed lines delineate borders of attached neuralized EB clusters. Scale bar, 50µm. (D) Representative Western blot in WT and TSC2−/− cultures at days 0, 1, 3, 5, 8, and 12 of monolayer-NPC differentiation in the absence (left panel) and presence (right panel) of 20nM rapamycin initiating 6h prior to differentiation. Proteins assessed are TSC2, ULK1, and P-ULK1 at Serine-757 (mTORC1) (left). Quantification by densitometry of 1 representative Western blot per cell line (H1, H9, 168), displayed as TSC2−/− over WT fold-change, for P-ULK1 at Serine-757. (E) Quantification as in panel S6D for P-ULK1 at Serine-555. (F) Quantification as in panel S6D for total LC3ß signal, LC3ß-I, LC3ß-II, and the LC3ß-II/ LC3ß-I ratio. (G) Western blot image as described in panel S6D during monolayer-NPC differentiation. Proteins assessed are mTOR, P-mTOR at Serine-2448, S6K, P-S6K at Threonine-389, AMPK, and GAPDH. (H) Representative image of Proteostat staining in WT and TSC2−/−NPC maintenance cultures at passage 6. Scale bar, 50µm.
(A) Representative images of c-H2DCFDA staining for reactive oxygen species (ROS) in NPCs and undifferentiated hPSCs; quantification of ROS levels in maintenance NPCs. Values are mean ± SEM (n=3, 1x each H7, H9, 168). (B&C) Cell number in dish at end-point of extracellular flux assays relative to WT in NPC and NCC (C) cultures, measured as number of Hoechst+ objects. Values are mean ± SEM (n=16 WT, n=10 and 11 TSC2−/−). (D-F) Toxicity observed in neural progenitor cells (left) and neural crest cells (right) 24 hours after treatments with Bortezomib (D), Carfilzomib (E) and Rapamycin or Chloroquine (F). Values are the mean ± SEM. (D-E)) n=7 for NPCs; 1x for H7, 4x for H9 and 2x for 168. n=11 for NCCs; 2x for H1, 2x for H7, 3x for H9 and 4x for 168. (F) n=4 for NPCs; 1x for H7 and 168, 2x for H9. n=4 for NCCs; 1x for H7 and H9, 2x for H1 and 168. Statistical significance (p < 0.05) was established by Two-way ANOVA and Tukey’s post hoc analysis. (G) Bliss independence analysis of interactions between the proteasome inhibitors (Bortezomib and Carfilzomib), Rapamycin and Chloroquine after 48h of exposure. Values are the mean degree of synergy (difference between theoretical and observed toxicities) ± SE. n=4 for NPCs; 1x for H7 and 168, 2x for H9. n=6 for NCCs; 1x for H7 and H9, 2x for H1 and 168)). Statistical significance (p < 0.05) was determined by the Holm-Sidak method.
Supplemental Multimedia
Video S1: Time lapse video of motility assay featuring WT and TSC2−/−NCCs with and without 100nM Rapamycin.
Sequencing of TSC2 exon 3 target region of TSC2−/− hPSC cell lines
Sanger sequencing was performed on all TSC2−/− hPSC cell lines to ensure proper integration of the ‘stop-codon’ donor sequence (highlighted in chromatograms).
AAVS1 Donor Plasmid Sequence
To aid in live cell assays and any potential in vivo studies, a CMV early enhancer/chicken β actin (CAG) promoter driven mCherry/Zeocin-resistance transgene was integrated into the AVVS1 locus. All cell lines utilized in this study, excluding 168 TSC2−/− cells, possess this transgene integration.
AAVS1: CAG-mCherry/T2A/Zeocin-Resistance donor plasmid sequence (in pUC19 backbone) and annotated map:
Annotated map:
Off-Target Mutation Analysis
Off target loci for each gRNA sequence utilized in this study were identified using the CRISPOR algorithm (77). The 3 highest scoring off target loci falling within an intron or exon were selected for PCR amplification of the off-target region followed by amplicon sequencing to identify potential off target mutations. No off-target cleavage induced indel mutations were detected across all CRISPR/Cas9 genome edited cell lines in this study.
Acknowledgements:
We thank Catherine Lawrence for her inspiration, members of the Stanford Lab for thoughtful review of this paper, and the Ottawa Bioinformatics and Human Pluripotent Stem Cell Core facilities. We thank the Tier 1 Canada Research Chair Program in Integrative Stem Cell Biology (WLS) for support. This research was supported by grants to WLS from the Canadian Institutes for Health Research (CIHR) (FRN-153188), the McEwen Centre for Regenerative Medicine (a Special Accelerated Discovery grant supported by Green Eggs and LAM), the LAM Foundation (Pilot grant LAM0123P01-17), and the United States Department of Defense via the Tuberous Sclerosis Complex Research Program of the Congressionally Directed Medical Research Program (W81XWH-14-1-0434). SPD was supported by the Judith R. Raymond Scholarship in Cancer Research. LMJ was supported by CIHR Banting and Ontario Institute of Regenerative Medicine Postdoctoral Fellowships and Cancer Research Society Next Generation of Scientists Scholarship.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.
- 17.↵
- 18.↵
- 19.
- 20.
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.
- 31.
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.
- 45.↵
- 46.↵
- 47.
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.
- 57.↵
- 58.↵
- 59.↵
- 60.
- 61.
- 62.
- 63.
- 64.↵
- 65.↵
- 66.
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
